PandaX: a liquid xenon dark matter experiment at CJPL

SCIENCE CHINA Physics, Mechanics & Astronomy • Article •

August 2014 Vol. 57 No. 8: 1476–1494

Progress of Projects Supported by NSFC

doi: 10.1007/s11433-014-5521-2

PandaX: a liquid xenon dark matter experiment at CJPL† CAO XiGuang4, CHEN Xun1, CHEN YunHua8, CUI XiangYi1, FANG DeQing4, FU ChangBo1, GIBONI Karl L.1, GONG HaoWei1, GUO GuoDong1, HE Ming1, HU Jie1, HUANG XingTao3, JI XiangDong1,6,7*, JU YongLin2, LI ShaoLi1, LIN Qing1, LIU HuaXuan2, LIU JiangLai1, LIU Xiang1, LORENZON Wolfgang5, MA YuGang4, MAO YaJun6, NI KaiXuan1, PUSHKIN Kirill1,5, REN XiangXiang3, SCHUBNELL Michael5, SHEN ManBing8, SHI YuJie1, STEPHENSON Scott5, TAN AnDi7, TARLÉ Greg5, WANG HongWei4, WANG JiMing8, WANG Meng3, WANG XuMing1, WANG Zhou2, WEI YueHuan1, WU ShiYong8, XIAO MengJiao1, XIAO Xiang1, XIE PengWei1, YE Tao1, YOU YingHui8, ZEN XiongHui8, ZHANG Hua2, ZHANG Tao1, ZHAO HaiYing1, ZHAO Li1, ZHOU XiaoPeng6 & ZHU ZhongHua8 1

The Institute of Nuclear and Particle Physics, Astronomy and Cosmology (INPAC) and Department of Physics and Astronomy, Shanghai Jiao Tong University, Shanghai 200240, China; 2 School of Mechanical Engineering, Shanghai Jiao Tong University, Shanghai 200240, China; 3 School of Physics and Key Laboratory of Particle Physics and Particle Irradiation (MOE), Shandong University, Jinan 250100, China; 4 Shanghai Institute of Applied Physics, Shanghai 201800, China; 5 Department of Physics, University of Michigan, Ann Arbor, MI, 48109, USA; 6 School of Physics, Peking University, Beijing 100080, China; 7 Department of Physics, University of Maryland, College Park, MD, 20742, USA; 8 Yalong River Hydropower Development Company, Ltd., Chengdu 610051, China Received May 13, 2014; accepted May 27, 2014; published online June 9, 2014

PandaX is a large liquid-xenon detector experiment usable for direct dark-matter detection and 136Xe double-beta decay search. The central vessel was designed to accommodate a staged target volume increase from initially 120 kg (stage I) to 0.5 t (stage II) and eventually to a multi-ton scale. The experiment is located in the Jinping Deep-Underground Laboratory in Sichuan, China. The detector operates in dual-phase mode, allowing detection of both prompt scintillation, and ionization charge through proportional scintillation. In this paper a detailed description of the stage I detector design and performance as well as results established during the commissioning phase are presented. dark matter, liquid xenon detector, underground experiment, time projection chamber PACS number(s): 95.35.+d, 14.80.Ly, 29.40.-n, 95.55.Vj Citation:

Cao X G, Chen X, Chen Y H, et al. PandaX: a liquid xenon dark matter experiment at CJPL. Sci China-Phys Mech Astron, 2014, 57: 14761494, doi: 10.1007/s11433-014-5521-2

*Corresponding author (email: [email protected]; [email protected]) †Contributed by JI XiangDong (Associate Editor-in-Chief) © Science China Press and Springer-Verlag Berlin Heidelberg 2014

phys.scichina.com

link.springer.com

Cao X G, et al.

Sci China-Phys Mech Astron

1 Introduction Overwhelming evidence from astrophysical observations indicates that only 20% of the matter in the universe is made of ordinary matter, while the remaining 80% is made of some form of matter coined “dark matter” [1]. Although these indirect astrophysical observations convince us that dark matter exists, dark matter has not yet been directly observed. The standard model of particle physics, which has been very successful in explaining the properties of ordinary matter, can neither explain dark matter’s existence nor its properties. Yet the discovery and identification of dark matter would have a profound impact on cosmology, astronomy, and particle physics. A leading dark matter candidate consistent with all astrophysical data is a weakly interacting massive particle (WIMP) [2–4]. WIMPs could be studied in standard particle physics through either observations of ordinary matter particles produced through DM annihilations in the halo of the Milky Way, production of DM particles through high-energy collisions in accelerators such as the Large Hadron Collider (LHC), or WIMPs could be detected through their interactions with atomic nuclei in specially designed detectors. Since the collision rates are expected to be very small, large detectors with low background rates and excellent detection capability for rare collisions are required for the detection of dark matter. There are a number direct detection experiments deployed in underground laboratories around the world. A recent review describing some of these experiments can be found in ref. [5]. When WIMPs scatter with atoms in a detection medium, they will recoil and generate kinetic motion of atoms (heat), ionization (free electrons) and scintillation (de-excitation of excited electrons). Direct detection experiments measure one or two or even possibly three of these signatures, depending on the choice of material. In pure semiconductors, such as those used by the CDMS collaboration [6] and others, one typically measures the electric current generated by electrons as well as hole carriers. In the case of noble liquid detectors (XENON100 [7], LUX [8], and others [9]), a light signal is usually measured by photo multiplier tubes (PMTs); ionization electrons drifting in an external electric field are either detected through their charge or through electroluminescence. In crystals, the light intensity is usually the only signal measured. In fact, the DAMA/LIBRA experiment measured scintillation light only [10]. For heat measurements, the detector has to be kept at very low temperature, typically at tens of milli Kelvin, which is a cryogenic challenge, particularly for large masses. Among all the direct detection experiments, the xenon dual-phase technology appears to be particularly promising. In fact, for the last 3–4 years, the XENON100 and LUX experiments, both using liquid xenon (LXe) have produced the best limits over a wide range of WIMP masses [7,8].

August (2014) Vol. 57 No. 8

1477

There are many reasons why LXe appears to be a good choice. First, the detection of both prompt scintillation and ionization electrons in dual-phase mode allows not only discrimination between nuclear and electron recoils, but also a fiducialization of events through the time projection chamber (TPC) technology. Second, xenon does not have long-lived radioactive isotopes and can be highly purified. Third, xenon has a large atomic mass, which entails a large WIMP scattering cross section. WIMP-nucleus scattering is coherent and hence the cross section is proportional to the square of the atomic mass number A. Moreover, xenon has a large Z which improves self-shielding from external gamma rays. Finally, xenon is not prohibitively expensive, allowing detector target masses to reach ton-scale within reasonable cost. Xenon liquefaction temperature is around 100°C, and thus cryogenics is relatively easy to manage. A crucial property of xenon as a WIMP detector is its outstanding background discrimination. A particle interacting in LXe produces both xenon excitation states and electron-ion pairs. The decay of excited states to the ground state results in scintillation light (S1) at a vacuum UV wavelength of about 175 nm. In the absence of a strong electric field, the recombination of electron-ion pairs will form excited states and produce additional scintillation light. If an electric field is present, the electron-ion recombination will be suppressed, and the free ionization electrons will drift in the liquid towards the anode, where they are extracted into the gas phase by a high electric field (≈ 10 kV/cm). The ionization signal is detected via its electroluminescence signal (S2) in the gas phase. Background discrimination explores the fact that the nuclear recoil (NR) signal from a WIMP elastic scattering event in LXe differs from that of electron recoils (ER). Most of the energy of the nuclear recoil is transferred to atomic motion and can not be detected, leaving only about 10%–20% observable energy relative to electron recoils of the same energy. The ionization density for nuclear recoils is much higher than that for electron recoils and therefore more electron-ion recombination takes place for nuclear recoils. This leads to a smaller ratio of ionization/scintillation (S2/S1), and provides 99.9% ER background discrimination. The Particle and Astrophysical Xenon (PandaX) collaboration was established in 2009 and was first supported by the Ministry of Science and Technology in China through a 973-project and by the Ministry of Education through a 985III-project at Shanghai Jiao Tong University (SJTU). The initial collaboration consisted mainly of physicists from SJTU, Shandong University and the Shanghai Institute of Applied Physics, Chinese Academy of Sciences. The collaboration was later joined by groups from the University of Maryland, Peking University, and the University of Michigan in 2011, and by the Yalong River Hydropower Development Co. Ltd. in 2012. The collaboration has received support from the National Natural Science Foundation of China (NSFC), and from some of the collaborating institu-

1478

Cao X G, et al.


tions. The PandaX experiment employs a liquid-xenon detector system suitable for both direct dark-matter detection and 136 Xe double-beta decay search. Similar to the XENON and LUX experiments, the PandaX detector operates in dual-phase mode, allowing detection of both prompt scintillation and ionizations through proportional scintillation. The central time projection chamber was designed to be expandable in stages, with the first stage accommodating a target mass of about 120 kg, similar to that of XENON100. In stage II, the target mass will be increased to about 0.5 t. In the final stage, the detector can be upgraded to a multi-ton target mass. Most sub-systems and the stage I TPC were developed in the particle physics laboratory at SJTU, and have been transported to the China Jinping Deep-Underground Laboratory (CJPL) in August 2012. After successful installation, two engineering runs were carried out in 2013. The system entered commissioning in December 2013 and has been collecting science data since late March 2014. A small prototype for PandaX was developed and is running in the particle physics laboratory at SJTU [11]. The underground lab, CJPL, emerged from a government-led project to construct two large hydropower plants next to and through the Jinping mountain, Sichuan, China [12]. Jinping is located about 500 km southwest of Chengdu, the capital of Sichuan province. It can be accessed either by car from Chengdu, or by a short flight to Xichang, followed by a 1.5 h car ride. The laboratory will be discussed in sect. 5 and a detailed description of CJPL can be found in ref. [13]. In this paper, we describe the goals and the technical realization of the PandaX detector system. In sect. 2 we describe the cryogenic system to liquefy, purify and maintain a ton-scale xenon detector. In sect. 3 we consider the design and construction of the stage I TPC. In sect. 4 we discuss the properties and performance of the photomultiplier tube system. In sect. 5 we discuss the sources of background for the experiment, including cosmic rays, the shield for environmental neutrons and gamma rays, and the xenon distilla-


tion system. We also briefly discuss the background simulations. In sect. 6 we describe the data taking electronics and data acquisition system (DAQ). We conclude the report with sect. 7.

2 Cryogenic and gas-handling system A reliable cryogenic and gas handling system is crucial for any noble-liquid detector system. As the size of the detector increases, cryogenic stability, safety, and efficiency of gas handling and purification become increasingly complex. For PandaX the goal was to build a system capable of storing, circulating and cooling xenon gas up to 2–3 t. So far the system has been set up and tested with 500 kg xenon. A detailed description of the cryogenics system was published in ref. [14]. A schematic sketch of the cryogenic and gas handling systems is shown in Figure 1 (left panel). The detector is enclosed in a double-walled cryostat for thermal insulation. Additionally, seven layers of aluminized Mylar foil are used in the insulating vacuum to reduce the heat load into the cryogenic detector. The inner vessel is 0.75 m in diameter and 1.25 m high. It is over-dimensioned for stage I to accommodate the future upgrade to stage II. It contains appropriate filler material to reduce the amount of xenon needed during operation. The outer vessel is made of 5 cm high-purity oxygen-free copper. The cryogenic system was designed as a series of independent modules, each with a specific function. The modules are connected to the detector via common tubing used by all modules not unlike the bus structure in a computer system connecting all devices. This modular cryogenic system was coined Cooling Bus (Figure 1, right). Each module performs a separate function such as evacuation, heat exchange, condensation, emergency cooling and sensor mounting. The gas handling system consists of a gas storage and recovery system, a circulation pump, a purification

Figure 1 Left panel: Schematic layout of the cryogenic and the gas system for PandaX. Right panel: Photograph of the Cooling Bus for liquefaction and recirculation. In the foreground the getter for Xe purification can be seen.

Cao X G, et al.


getter and a heat exchanger. The detector and the Cooling Bus are connected by concentric tubes to the evacuation module consisting of a fore-pump and a turbo pump for the inner vessel, and a pair of pumps and a zeolite filled cryo pump for the outer vessel. Normally the vacuum jacket is maintained by the fore-pump and the turbo pump for the outer vessel. A cryo-sorption pump is also connected in order to provide continued pumping of the outer vessel in case of power failures. The system is cooled by a single Iwatani PC150 pulse tube refrigerator (PTR), driven by a 7.3 kW air-cooled M600 compressor from Oxford Instruments. Its cooling power was measured to be around 180 W [14]. It is mounted on a cylindrical copper block made of the oxygen-free copper. The copper block closes off the inner chamber and acts as a cold-finger for liquefying the xenon gas. The PTR can thus be serviced or replaced without exposing the detector volume to air. A copper cup that serves as a heater is installed between the PTR cold-head and the cold-finger. The temperatures of the cold-head and cold-finger are measured by Pt100 temperature sensors. A PID temperature controller (Lakeshore 340) regulates the heating power to keep the temperature of the cold-finger constant. In case of failure of the PTR system, e.g., during power failure, a pressure sensor will start the flow of LN2 through a cooling coil above a pressure set point about 0.5 bar above the normal operating pressure. This cooling will continue until the xenon pressure drops below a second set point, about 0.5 bar below the normal operating pressure. The pressure sensor and the LN2 control valve are powered by a dedicated uninterruptible power supply (UPS). The warming up and LN2 cooling are shown by their effect on the xenon pressure in Figure 2. Note that the emergency cooling system is always operational, but normally does not contribute to the detector cooling since the set points are never reached. This cooling can continue indefinitely if sufficient LN2 is available. As ultimate safety device a burst disc limits the maximum pressure in the detector. The burst pressure with 3.5 barA is sufficiently above the emergency set point not to trigger when LN2 cooling is available. The pressure, however, is low compared with the maximum allowable pressure for the PMTs. Since the burst pressure is rather critical for safeguarding the xenon and the PMTs, a burst disc with a precisely controlled pressure rating of ±5% was chosen. A heat-exchange module is used as part of the purifica-

Figure 2


tion system which is designed to purify the liquid xenon continuously. Constantly recirculating the xenon through a high temperature getter (SAES, PS4-M750-R-2, Max: 100 SLPM) gradually improves purity and removes electro-negative molecules originating from out-gassing of the surfaces of the detector materials. The impurity level in LXe determines the attenuation length for scintillation light and the life time for drifting charges. The most common electronegative impurities are H2O, O2, CO2, and CO. The gas recirculation system is driven by a double diaphragm pump (KNF, PM26937-1400.12) and/or a custom-made Q-drive1). The two pumps are installed in parallel, providing redundency. The flow rate in the current operation is about 30 SLPM. The setup is sufficiently powerful to purify xenon gas in a ton-scale detector. For ton-scale operation, hundreds of thousand of liters of xenon gas has to be recovered and stored at room temperature. At Jinping lab, custom-made 220 L steel high-pressure bottles and LN2 dewars are used for this purpose. Their working pressure is 8 MPa, with each dewar storing about 250 kg xenon gas. These bottles can be cooled down by filling LN2 into the dewars to recover xenon gas from the detector. Tests showed that it takes 2–3 d to recover about 500 kg xenon in the detector for the stage I experiment. To control the LXe level in the inner vessel, a volume of about 10 L, called the overflow chamber, is used. The LXe can flow through a pipe from the TPC to the overflow chamber. A Bowden Cable is attached to the end of the pipe and the height of the pipe’s outlet can be tuned from the outside. With this method, the liquid level in the TPC can be controlled with a precision of 0.1 mm. The liquid flowing to the overflow chamber is recirculated through the getter and liquefied back into the inner vessel. The cryogenic system for the PandaX LXe detector was tested extensively in the SJTU particle physics lab in 2012. Since moving it to CJPL in August 2012, two engineering runs were performed. The system was filled with 450 kg of xenon, which was later recovered. It takes approximately 3 d to liquefy the xenon gas with LN2 assisted cooling. Each time the system performed as expected. Figure 3 shows typical values for inner pressure, outer vacuum level, and cold finger temperature as a function of time as the system is running. These values can be accessed through a slowcontrol system via the internet. They indicate stable running conditions.

Results of an emergency cooling system test: Inner pressure as a function of time when the electric power is off.

1) Wang H G. Private communication

1479

1480

Cao X G, et al.

Figure 3


Inner pressure, outer vacuum level and temperature of the cold-finger vs time, with 450 kg liquid xenon in the inner vessel.

3 PandaX stage I TPC The time projection chamber design for the PandaX stage I experiment focussed on maximizing light collection efficiency to achieve a low energy threshold and a high sensitivity to light dark matter at around 10 GeV/c2. With a field cage diameter of 60 cm and a drift length of 15 cm a liquid xenon target mass of 120 kg can be accommodated. By increasing the drift length to 60 cm, the much larger volume required for stage II can easily be achieved while preserving the overall design. The mechanical design for the TPC is shown in Figure 4. It consists of three major parts: top PMT array, field cage, and bottom PMT array. In the following the TPC construction will be discussed in detail. 3.1


Field cage

The field cage, shown in Figure 5, contains a teflon cylinder of 15 cm height and 60 cm diameter. The teflon cylinder has 36 pieces of 5-mm thick teflon panels interlocking with adjacent panels. Among them, 18 pieces of teflon panels are designed with 5-mm thick tablets on their top and bottom ends, by which each of them are fixed on a teflon support by teflon bolts. Those 18 pieces of teflon supports serve as the bearing carrier of the field cage and hold all other components besides the teflon panels. There are altogether 4

electrodes named anode, gate grid, cathode and screening electrodes from top to bottom in the TPC. Each electrode has an inner diameter of 60 cm, which is the same as that of the field cage. To fabricate the gate grid, cathode and screening electrodes, stainless steel wires (304 or 316 L) with 200 m diameter are pressed between two stainless steel rings hold by fasteners (see Figure 5 center). The rings are made of 316 L stainless steel with 3 mm thickness and 15 mm width. The spacing between two wires is 5 mm, which results in 96% optical transparency for the electrode. Each wire has a tension of 2.8 N (43% of yield strength), provided by 288 g weight rods hanging on both side of each wire during production. The anode electrode is made of photo-etched mesh with crossing bars of 200 m width and 5 mm spacing, providing 92% optical transparency. The anode mesh is fixed above the grid with 5-mm thick teflon rings (see Figure 5 right). Thus the distance between the anode mesh and the gate grid wires is 8 mm. Anode, gate grid and cathode electrodes are fixed on the teflon supports as shown in Figure 5. The teflon supports are mounted to the top copper plate by PEEK bolts so the field cage is integrated with the top PMT array. For drift field uniformity, 14 pieces of OFHC copper shaping rings were arranged outside the teflon panels, between the gate grid and the cathode in equal spacing. They are constructed of OFHC copper tubes with 6 mm outer diameter and 5 mm

Figure 4 Mechanical design of the stage I TPC. Left panel: Full view showing the completely integrated TPC. Right panel: Cross-sectional view showing the detailed components.

Cao X G, et al.



1481

Figure 5 Left panel: Mechanical design of the field cage showing the teflon panels and supports, three electrodes and the field shaping rings. Center panel: Photograph of the wire electrode for the gate grid, the cathode and the screening electrodes. Right panel: Photograph showing the anode mesh is fixed above the grid ring with a teflon ring in between.

Figure 6 Photographs showing the resistors between the shaping rings, the gate grid and the cathode (left), a resistor mounted between anode and the gate grid (center), and the field cage, together with the top PMT array, hanging from the top flange of the inner vessel (right).

inner diameter, and clamped by teflon supports and teflon panels (see Figure 6). 500 M surface-mount resistors (SM20D from Japan FineChem Company, Inc.), rated for a maximum voltage of 5 kV, are tied by bare copper wires between each two adjacent shaping rings, between shaping rings and cathode or gate grid, and between gate grid and anode (see Figure 6). Two resistor chains are mounted on the electrodes to prevent disruption of the experiment in case of single resistor failure during operation. A photograph showing the field cage, integrated with the top PMT array, installed in the detector is shown in Figure 6. 3.2

PMT arrays

The 143 top PMTs (Hamamatsu R8520-406) are mounted to an 8-mm thick oxygen-free high-conductive (OFHC) copper plate. Each top PMT is attached to a base by a pin socket connector. The base in turn is mounted to the OFHC copper plate with spring-loaded bolts. The 143 PMTs are arranged uniformly in 6 concentric rings around a center PMT. The spacing of adjacent rings is 52.5 mm and the number of PMTs in each ring is 36, 36, 28, 20, 14 and 8 from edge to center. To achieve good position reconstruction for events at large radius, the diameter of the outermost PMT ring is

630 mm, which is larger than that of the field cage. The teflon reflector covering the space between PMTs is made of five pieces of 6-mm thick teflon plate with openings according to the arrangement of PMTs. Those teflon plates are “interlocking” with each other to avoid gaps from appearing due to the shrinkage of teflon at LXe temperature, and are mounted on the OFHC copper plate by PEEK bolts. The 37 bottom PMTs (Hamamatsu R11410-MOD) are supported by a 8 mm thick OFHC copper plate. Unlike for the top PMTs, each bottom PMT is held by a pair of stainless steel clamps, fixed in between two 8 mm thick teflon clamps which are attached to the OFHC copper plate by stainless steel bolts. The bottom PMT bases are attached to each PMT by pin-socket connection directly, as shown in Figure 7. Teflon reflectors are mounted between the PMT windows to increase light collection. The teflon reflectors are made up of 7 pieces of 2.4 mm thick teflon sheets, which are attached to the OFHC copper plate by teflon bolts. To reduce the electric field strength near the bottom PMT photocathode to acceptable levels, the grounded screening electrode wires are at 5 mm above bottom PMT windows. Similar to in the field cage, 36 pieces of shorter teflon panels are designed to improve light collection efficiency, and fixed on the OFHC copper plate by sharing the same teflon

1482

Cao X G, et al.



Figure 7 Left panel: Mechanical design of the bottom PMT array and the teflon reflectors. Center panel: Photograph showing all 37 R11410 PMTs installed on the copper plate of the bottom PMT array. Right panel: Photograph showing the bottom PMTs with the copper filler and how they are connected to the bottom feedthroughs.

bolts with the screening electrode. The entire array with all PMTs assembled on the copper plate is shown in Figure 7. The PMT HV and signal cables are connected on the bottom flange of the inner vessel. To minimize the empty space below the PMTs and to shield gamma rays from the bottom flange, a copper filler is installed below the PMT array (see Figure 7). 3.3

Initial operation of the TPC

For a dual phase operation of the PandaX TPC, the anode is connected to the ground, while the gate and cathode grids are connected with negative high voltages (HV). The cathode is connected to a custom-made HV feedthrough, while the gate grid is connected to a commercial feedthrough rated up to 10 kV. During the engineering run in liquid xenon with no PMT signal readout, the high voltage on the cathode reached 36 kV before electrical breakdown, and the high voltage on the gate grid reached 6 kV (the maximum value of the power supply). However, during the full operation of the detector with PMT readout, micro-discharge signals were observed by the PMTs when the cathode voltage reaches 20 kV. Similar discharge signals were observed when the gate grid is above 5 kV. These discharges produce many small signals at the single or a few photoelectrons level, preventing the useful data taking. Thus it sets the limitation of the drift field at 1 kV/cm across the 15 cm drift gap. To extract electrons from the liquid to the gas phase, we set the liquid level in between the anode and the gate grids. The liquid level can be adjusted with an overflow point controlled by an external motion-feedthrough. The level should be set at least 3 mm above the gate wires so that it covers the 3 mm-thick stainless steel ring to avoid discharges from any sharp point on the ring if it’s exposed in the gas. We set the level at 4 mm above the gate grid for the initial operation of the TPC, giving a 4 mm gas gap. For the extraction field in the gas above the liquid xenon, a 5 kV voltage between the anode and gate grids provides an extraction field of 8.3 kV/cm, corresponding to an electron extraction yield of 90% according to ref. [15]. Following the adjustment of high voltages on the cathode

and gate grids, both S1 and S2 signals are observed by the PMTs after the liquid xenon reached a good purity. A typical S1-S2 waveform summed from all PMTs and their S2 signal distribution among the PMTs for a single-site event is shown in Figure 8.

4 Photomultiplier system PandaX uses specially-developed PMTs to detect prompt and proportional scintillation light. The photomultiplier system has to satisfy many requirements, such as good quantum efficiency to VUV light (178 nm) from xenon, low radioactivity, single photoelectron (SPE) resolution, good timing resolution, cryogenic operation (100°C) suitability, high-pressure operation (>3 atm) suitability, and minimal outgassing from the bases and cables. These requirements are met by the one-inch Hamamatsu R8520-406 and by the three-inch Hamamatsu R11410-MOD photomultiplier tubes, which instrument the top and bottom photomultiplier arrays, respectively. In this section, we describe various aspects of the PMT system, including the basic properties, bases, high voltage and decoupler, feedthrough and cabling, calibration and test results. 4.1

Photomultiplier tubes

The Hamamatsu model R8520-406 is a 10-stage, one-inch square photomultiplier tube, rated for a temperature range of 110°C to +50°C, and 5-atm pressure resistance. The typical gain is 106 at a HV setting of 800 V. The cathode window is made of Synthetic Silica, and the cathode material is Bialkali, yielding an excellent quantum efficiency of about 30% at 175 nm. The radioactivity of this tube has been measured by XENON100 collaboration in ref. [16]. The Hamamatsu model R11410-MOD with ceramic stem is a 12-stage, three-inch circular photomultiplier tube, also rated for LXe temperature with a typical gain of 5×106 at 1500 V. The maximum pressure rating, updated in 2011, is 0.4 MPa (absolute). The quantum efficiency is >30% at 175 nm. Radiopurity measurements were performed elsewhere [16], as well as in our own counting station at CJPL (see

Cao X G, et al.

Figure 8


1483

The summed waveform and S2 signal distribution on all PMTs of a typical single-site event during the calibration run.

sect. 5.3). Note that for low temperature operation, flying leads were cut off to a length of 10 mm from the ceramic stem for attachment to the voltage divider. 4.2


PMT voltage dividers

Positive HV voltage dividers (bases) are chosen to put the photocathode at ground a) to reduce the noise level and improve the single photoelectron (SPE) resolution, b) to eliminate potential interference with the electric field profile of the TPC, and c) to allow bringing signal and HV in/out the vessel through the same coaxial cable (decouplers will be placed outside the vessel) to minimize the number of cable feedthroughs. The design of the voltage divider for the R8520-406 PMT follows recommendations from Hamamatsu, with schematics shown in Figure 9(a). To limit the heat load for cryogenic operation, the total resistance of the divider chain is 13 M, so that heat power from each base is only 0.05 W under normal voltage setting (800 V). The base is backterminated with 100 k resistor R16, which increases not only the charge collection at the output end but also the low frequency band width, which is critical to minimize signal distortion for S2-type signals. With the rest of the frontend electronics terminated at 50 , this results in some signal reflection (see sect. 4.6). A 10 nF capacitor C4 between the cathode and anode is necessary to remove signal oscillation. The design of the R11410-MOD base is very similar to that of the R8520-406 base, with its schematic shown in Figure 9(b). For all base capacitors, low background ceramic X7R capacitors from Kyocera Inc., rated at 1 kV, were selected. Two capacitors were put in series between the cathode and the anode for the R11410-MOD base. To

reduce radioactivity, C2 and C3 were removed from both bases. Signal distortion of typical S2 pulses were measured to be negligible (see sect. 4.6). A photograph of an R8520-406 base is shown in Figure 10. Several technical issues were addressed in its construction. Cirlex, a kapton-based material, was chosen as the base material due to its good radiopurity and low outgassing characteristics. Pure silver tracks were deposited onto the PCB without other add-ons. Ceramic capacitors from Kyocera, and lead-free soldering tin (Sil-Fos) from LucasMilhaupt/Handy & Harman were selected and used on the base to minimize radioactivity. KAP3, a UHV coaxial cable from MDC Inc was selected as the signal/HV cable. The receptacles for the PMT pins were chosen from Mil-Max Inc. The base of the R11410-MOD PMT was constructed in a very similar way, as shown in the right panel of Figure 10. The PMTs and bases were radioassayed in the HPGe counting station in Jinping, with results summarized in sect. 5.3. The radioactivity levels from the PMTs are in agreement with those reported in ref. [16]. The high 238U/232Th/ 40 K content in the one-inch base is likely due to the particular type of pin receptacles used, since other material components are identical to those on the three-inch bases. 4.3

Signal-HV decoupler and high voltage system

The signal (fast pulses) and the DC high voltage are decoupled outside the detector via a decoupler module. A similar design from the Daya Bay experiment [17] was followed. A schematic sketch of the decoupler is shown in Figure 11 (left). The decoupling capacitor, rated for 2 kV, is chosen to be 100 nF based on a SPICE simulation and bench tests to minimize the signal distortion for S2-like signals. At the

1484

Figure 9

Figure 10

Cao X G, et al.



PMT base schematics for (a) the R8520-406 and (b) the R11410-MOD bases.

Photographs of the PMT bases for (a) the R8520-406 and (b) the R11410-MOD models.

input of the DC high voltage, a 3-stage high pass filter is implemented to remove the ripples from the high voltage supply. A photograph of a 48-channel decoupler module (4 U) is shown in Figure 11 (right). Each module consists of four 12-channel PCBs. The high voltage inputs into the module are through four China-standard mil-spec DB25 connectors, tested for 2 kV. Cables leading to PMTs are through 48 SHV connectors on the back panel, and the decoupled signals to the electronics are through the 48 BNC connectors on the front panel. The PMT high voltage system is from CAEN SpA, with a SY1527LC [18] main frame and four A1932AP [19] modules, each supplying 48 HV channels up to 3 kV. The

ground of these HV channels is configured as float from the ground of the main frame. The output connector on each module is a 52-pin HV connector by Radiall. A custom fan-out cable is made to connect each Radiall connector to four DB-25 connectors, as the side input to the decoupler box (see Figure 11). To reduce the 200 kHz noise from the HV supply, additional RC filters were implemented before the HV enters into the decoupler box. 4.4

Cabling

The PandaX PMT cabling scheme is shown in Figure 12. As mentioned previoulsy, the KAP3 cables connect to indi-

Cao X G, et al.



1485

Figure 11

Left panel: Schematic diagram of a single channel signal-HV decoupler circuit. Right panel: Photograph of a 48-channel decoupler module.

Figure 12

Schematic overview of the PandaX cabling scheme.

vidual PMT bases. For the top PMT bases, the cables exit the inner vessel through six 48-pin double-sided high voltage CF35 feedthrough flanges by Kyocera. The lower PMT feedthroughs are two commercial double-ended 41-pin low temperature HV feedthroughs by MPF Products Inc. The connectors to these feedthroughs are custom-made with PEEK as insulator and lead-free sockets by TE/AMP. The RG316 coaxial cables carry the PMT signals/HV into the outer vessel vacuum. From the feedthrough flange on the inner vessel, every six cables are grouped with the other end soldered onto a male China-standard mil-spec DB15 connector, with core and ground individually separated. A custom cable assembly, consisting of 12 RG316 cables, connect two of these connectors to a 24-pin double-sided

vacuum feedthrough by LEMO Inc. Eight such LEMO feedthroughs are sealed against a ISO160 flange on the outer vessel via o-rings. For the bottom PMT cables, a cable assembly with RG316 cables connects the bottom feedthrough with another MPF double-sided 41-pin feedthrough mounted on the wall of the outer vessel. The cables outside the outer vessels are RG316 as well, each connecting to the decoupler modules through individual SHV connectors. The total cable length from the decoupler to the PMT base is approximately 5 m. 4.5

LED calibration system

A fiber optics system is installed in the detector to carry

1486

Cao X G, et al.


external LED light pulses into the detector for single photoelectron calibration. Optical fibers feed into the inner vessel through commercial ultra-high vacuum fiber feedthroughs. The open ends of the fibers were inserted into three teflon rods mounted on the wall of the TPC. Three external blue LEDs are driven by custom-built pulsers [20]. Fast light pulses (